Thursday, August 24, 2017

Interoperability and HealthIT: Are we there yet?

Are we there yet?  The short answer, as I quoted from a speaker earlier last week, is: "There is no done with this stuff".  The longer answer comes below.

If you are as old as I am, you remember having to have a case full of Word Perfect printer drivers, Centronics and Serial cables, and you might even have a Serial breakout box to help you work out problems setting up printers.  Been there, done that

What's happened since then?  Well, first we standardized port configurations based on the "IBM PC Standard".  Except that then we had to move to 9 pin serial cables.  And then USB.  And today, wireless.  Drivers were first distributed on disk, then diskette, then CD.  And now you can download them from the manufacturer, or your operating system will do that for you.

If you happen to have a printer that isn't supported, well, if it supports a standard like Postscript, we've got a default driver for that, and for PCL printers, and several dot matrix protocols.  So, today you can buy a printer, turn it on, autoconfigure it, and it just works, right?  Mac users had it a bit easier, but they still went from the old-style Mac universal cables to USB to ...

I upgraded my network infrastructure the other day, and come to find out my inkjet printer that had been working JUST fine on all the computers in the house, and iPhones and iPads, no longer worked on my various Apple devices.  I tracked it down to a compatibility issue between new features of my WiFi router and my old printer.  As a consumer, my expectations of interoperability were definitely NOT met.

Which brings us back to my main point.  The expectation of users with regard to interoperability still isn't being met, even if the situation is improving.  It took us twenty some years to get from where we were then to where we are now, and some configurations still aren't "Plug and Play" with respect to printing.

To figure out how to measure where we are with regard to interoperability, we first need to figure out what it is we want to measure.  And then we need to figure out how to measure the distance to that goal.  When "where we want to go" is an obscure location, figuring out how far we have to go is huge challenge.

Let's assume we want "Plug and Play" interoperability.  What does that actually mean?  We probably want to start with a basic platform and set of capabilities.  You have to define that, first functionally, and then in detail so that it can be implemented.  Then we have to talk about how things are going to connect to each other.  Connecting things (even wirelessly) is hard to do right.  Just ask anyone whose ever failed to connect their Bluetooth headset to their cell phone.  Do you have any clue how much firmware (software embedded in hardware) and software is necessary to do that right?  We've actually gotten that down to a commodity item at this stage.

If we look at the evolution of interoperability in hardware spaces such as the above, we can see a progression up the chain of interoperability.

1. Making a connection between components.
This is a progression from wires and switches to programmable interfaces to systems that can automate configuration of a collection of components.
2. Securing a connection over the same.
This is a progression from internal physical security, to technical implementations of electronic security, to better technical implementations, with progressions advancing as technology makes security both easier and harder depending on who owns it.
3. Authenticating/authorizing interconnected components.
We start from just establishing identities, to doing so securely, and from complex manual configurations, to more user friendly configurations, and finally to policy based acceptance.  At some point, some human still has to make a decision, but that's getting easier and easier to accomplish.
4. Integrating via common APIs or protocols.
Granularies start out at a gross level (e.g., CDA Document), and get more refined as time goes by and communication speed and response times get better, and drive from data (a set of bits) to functional ( function to produce a set of bits to understand) and back to data again (finer grained data) and algorithms (functional instructions again on how to produce data).  This is a never ending cycle.
5. Adapting to capabilities of connected components.
This starts at the level of try and see if it works and respond gracefully to errors, to declaration of optional feature sets, to negotiations between connected components about how they will work together.
6. Discovering things that one can connect to.
We first start by making a list for a component, then by pointing components to lists of things, then by pointing components to places where they can find pointers to lists, and finally, by broadcast protocols where basically, all you need to know is how to look around your environment.  Generally, there will always need to be a first place to look though (it might be a radio bandwidth, a multicast address, or a search engine location)
7. Intelligently interconnecting to the environment one is in.
The final destination.  We don't know what this really looks like for the most part.

Where we want to go is that final stage, and arguably, that's what we have finally begun to reach with the end user experience installing a printer (with some bobbles).  There's still some hardware limitations on Bluetooth devices because those are mostly small things, but even that is reached stage 6.  For healthcare, we are somewhere around stage 4 with FHIR.  CDS Hooks is arguably stage 5. Directories and networks like Carequality or Commonwell or Surescripts RLS will be progress towards stage 6.

The progression down this stack takes time, and the more complex the system, the longer it takes. Consider that printers, headsets and even cell-phones and laptops aren't enterprise class computing systems. The IT industry in general is making progress, but we aren't at a stage yet where enterprise level ERP, CRM and FMS systems are much further along than level 5 or 6, even multi-million dollar industries.  The enterprise level EHR and RCM and EDI systems used in similar sized businesses are moving a bit slower (a classic issue in HCIT).

So, back to measurement.  Are we there has a context.  If your goal is to get to stage 7, be prepared to wait a while and continue be frustrated.  In 2010, my family drove nearly 5000 miles to get sushi. There were plenty of stops along the way, and getting to each was exciting.  If you want to have fun along the journey, identify the way points, and make a point that this IS your NEXT destination. Otherwise, sushi is still a very long way off.

   -- Keith


Post a Comment